Year 7 of looking a multiple facets of AI technology, applications, research, and more - 212 DENSE google slides. (start with the blog post)
Published by Nathan Benaich on 10 October 2024.
If last year was the breakout moment for foundation models, this was the year of consolidation. While this year’s report continues to document significant advances in capabilities, our relationship with foundation models has changed. Researchers now have a significantly better understanding of how they can accelerate their work and how best to mitigate their disadvantages. Meanwhile, companies are now investing real effort in moving from merely building models to creating products.
In last year’s report, we asked whether generative AI products could hold onto users once the initial ‘wow’ factor (and the trial subscription) came to an end. That question has been answered definitively. OpenAI is now raking in billions of dollars in revenue, while the likes of ElevenLabs and Synthesia have become everyday tools for Fortune 500 companies.
Growing adoption, however, also means growing challenges. Some of these are on a policy level. The boom in international summits, pacts, and protocols has failed to paper over significant disagreements about governance. Big tech companies are at war with European regulators, while California’s proposed AI regulation fuelled a civil war in the community. The EU AI Act may now have passed into law, but there is a growing sense of buyers’ remorse on the continent.
For a long time, discussions of AI have been dominated by model scaling laws and their consequences. This year, companies have been forced to confront very real physical constraints, as their demands for power, water, and land place an increasing strain on computing infrastructure. Optimistic net zero commitments embraced idealistically five years ago now seem to be in serious jeopardy. At the same time, this AI infrastructure build out requires capital well beyond the means of many institutional investors, forcing companies to look overseas, with geopolitical implications.
Amid all these challenges, one undisputed winner has emerged. NVIDIA has joined the $3T club, become a stock market bellwether, and is arguably the most powerful company in the world. A growing number of challengers, restrictions on its China business, and belated software investment by its old rivals have failed to leave a scratch.
While NVIDIA is the most extreme example, public companies at the forefront of AI development have gained trillions of dollars in enterprise value. All the more impressively, they’ve done so at a time of high interest rates and wider market stagnation. Combined with growing adoption rates, huge infrastructure build-outs, and nuclear power plants being fired up just to service AI-related demand - it feels like we’re truly entering a new epoch.
Key takeways from the 2024 Report include:
- Frontier lab performance begins to converge and proprietary models lose their edge, as the gap between GPT-4 and the rest closes. OpenAI o1 put the lab back at the top of the charts - but for how long?
- Planning and reasoning take priority in LLM research, as companies explore combining LLMs with reinforcement learning, evolutionary algorithms, and self-improvement to unlock future agentic applications.
- Foundation models demonstrate their ability to break out of language, supporting multimodal research across mathematics, biology, genomics, the physical sciences, and neuroscience.
- US sanctions have limited effects on Chinese labs’ ability to produce capable models, as a combination of stockpiles, approved hardware, smuggling, and cloud access allow them to build highly performant (V)LLMs. Meanwhile, China’s efforts to build a domestic semiconductor industry remain scrambled.
- The enterprise value of AI companies has hit $9T, as public companies experience a bull market for AI exposure. Investment in private AI companies also increased, but by an order of magnitude less, despite GenAI megarounds in the US.
- A handful of AI companies begin to generate serious revenue, including foundation model builders and start-ups working on video and audio generation. However, as models get cheaper as part of the corporate land-grab, questions around long-term sustainability go unanswered.
- The pseudo-acquisition emerges as an off-ramp for AI companies, as some companies struggle to find a viable business model as staying at the frontier proves costly.
- The existential risk discourse has cooled off, especially following the abortive coup at OpenAI. However, researchers have continued to deepen our knowledge of potential model vulnerabilities and misuse, proposing potential fixes and safeguards.
The report is a team effort and we’re grateful to the AI community who continue to create the breakthroughs that power this report. Thank you to our reviewers who kept us honest.
We write this report to compile the most interesting things we’ve seen, with the aim of provoking an informed conversation about the state of AI. So, we would love to hear any thoughts on the report, your take on our predictions, or any contribution suggestions for next year’s edition.
Enjoy reading!
Nathan
🏡 Home 📧 Sign up to Air Street Press
Authored on the interwebs by:
This work is licensed under a Creative Commons Attribution 4.0 International License.